In the diagnosis of many forms of disease, as well as when following the effects of treatment, it would often be desirable to use labelled proteins that bind to specific target structures in the body. For example, when diagnosing or treating cancer, it would be desirable to be able to detect both primary tumours and metastases using labelled tumour-specific antibodies. Many reports have appeared on the labelling of proteins and antibodies by random chemical attack on their side chains. In such a process, most frequently, the side chains of the tyrosines are iodinated (Mach et al. Cancer Research 43, 5593–5600 [1983]), or the side-chains of the lysines are acylatea. In this latter case the acylation is often by groups that chelate metals (e.g. Hnatowich et al., Science 220, 613–615 [1983]). Subsequently, the chelating groups can be used to bind radioactive metals. It has also been suggested but not yet been satisfactorily tested to bind to such molecules paramagnetic ions for nuclear magnetic resonance (NMR) imaging (Brady et al., Magnetic Resonance in Medicine 1, 286 [1984)). The labelling of proteins, especially of antibodies, however, has so far always been effected in a more or less random way.
Random substitutions on biological active proteins, for example random substitutions on antibody molecules, can have number of drawbacks:    1. If by chance a particularly reactive site were to lie in the active site of the protein a substitution at this site would possibly inactivate the protein, e.g. a is particularly valuable monoclonal antibody might be rendered totally useless if by chance a side chain particularly reactive towards substitution were to lie in the antigen-binding site. The substitution would then inactivate the antibody.    2. Even when the active site of the protein (e.g. an antibody) escapes serious damage, a high number of substitutions on the protein—which may be desirable, e.g. in order to have a high intensity in case of radioactive labelling via chelating groups—might change its physico-chemical properties (e.g. solubility).    3. A random, multiple substituted product constitutes a heterogenous mixture of molecules with different properties, with attendant problems of assuring constant properties from batch to batch.